Unsupervised Artificial Intelligence Research Articles

Data-driven normative values based on generative manifold learning for quantitative MRI

In medicine, abnormalities in quantitative metrics such as the volume reduction of one brain region of an individual versus a control group are often provided as deviations from so-called normal values. These normative reference values are traditionally calculated based on the quantitative values from a control group, which can be adjusted for relevant clinical co-variables, such as age or sex. However, these average normative values do not take into account the globality of the available quantitative information. For example, quantitative analysis of T1-weighted magnetic resonance images based on anatomical structure segmentation frequently includes over 100 cerebral structures in the quantitative reports, and these tend to be analyzed separately. In this study, we propose a global approach to personalized normative values for each brain structure using an unsupervised Artificial Intelligence technique known as generative manifold learning. We test the potential benefit of these personalized normative values in comparison with the more traditional average normative values on a population of patients with drug-resistant epilepsy operated for focal cortical dysplasia, as well as on a supplementary healthy group and on patients with Alzheimer’s disease.

Biosecurity adoption in Québec dairy farms: results from a risk assessment questionnaire analyzed using conventional and unsupervised artificial intelligence methods

This study documents the current state of biosecurity on dairy farms in Québec following the implementation of a mandatory biosecurity risk evaluation that was part of the proAction® accreditation program developed by Dairy Farmers of Canada. Using a cross-sectional design, 3,825 risk assessment questionnaires completed between 2018 and 2021 were extracted from Vigil-Vet database, which is a software utilized by veterinarians for conducting the proAction® risk assessment. Descriptive statistics were used to summarize the practices adopted by dairy producers. Additionally, multiple correspondence analysis was used to explore the association between the diseases of most concern and the adoption of biosecurity practices. Moreover, we used a hierarchical cluster analysis on principal components to identify distinct patterns of biosecurity practices among dairy producers. This analysis enabled the identification of typologies or clusters of farms based on the specific biosecurity practices they currently employ. The results of the descriptive statistics indicated that mastitis was the disease of most concern for most dairy farmers (40%). Moreover, given that only 10% of the 2,237 dairy farmers who acquired animals adhered to quarantine practices, there seems to be a need for improved implementation of biosecurity measures aimed at restricting the introduction of diseases when introducing new animals. Conversely, cleaning stalls and health equipment were adequately addressed by 95% and 86% of dairy producers, respectively. The multiple correspondence analysis indicated no significant association between the disease of most concern and the farm's biosecurity profile, except for respondents who identified digital dermatitis as their disease of most concern. Through the hierarchical cluster analysis, 3 clusters were identified among 3,581 farms: (1) Cluster 1 included farms with good management of sick animals; (2) Cluster 2 included farms with good management of young animals; and (3) Cluster 3 included farms with poor management of sick animals and young animals. Our study makes an important contribution by providing valuable insights into the biosecurity practices currently adopted on Québec dairy farms. It establishes a baseline for assessing progress in biosecurity practices adoption and serves as a reference point for future evaluations. In addition, these findings play a key role in monitoring the effectiveness of interventions aimed at improving biosecurity on dairy farms. By making use of this knowledge, stakeholders can make informed decisions that prioritize animal health, increase productivity, and ensure sustainability of the dairy industry.

Unsupervised and supervised AI on molecular dynamics simulations reveals complex characteristics of HLA-A2-peptide immunogenicity.

Immunologic recognition of peptide antigens bound to class I major histocompatibility complex (MHC) molecules is essential to both novel immunotherapeutic development and human health at large. Current methods for predicting antigen peptide immunogenicity rely primarily on simple sequence representations, which allow for some understanding of immunogenic features but provide inadequate consideration of the full scale of molecular mechanisms tied to peptide recognition. We here characterize contributions that unsupervised and supervised artificial intelligence (AI) methods can make toward understanding and predicting MHC(HLA-A2)-peptide complex immunogenicity when applied to large ensembles of molecular dynamics simulations. We first show that an unsupervised AI method allows us to identify subtle features that drive immunogenicity differences between a cancer neoantigen and its wild-type peptide counterpart. Next, we demonstrate that a supervised AI method for class I MHC(HLA-A2)-peptide complex classification significantly outperforms a sequence model on small datasets corrected for trivial sequence correlations. Furthermore, we show that both unsupervised and supervised approaches reveal determinants of immunogenicity based on time-dependent molecular fluctuations and anchor position dynamics outside the MHC binding groove. We discuss implications of these structural and dynamic immunogenicity correlates for the induction of T cell responses and therapeutic T cell receptor design.

The Evolution of Artificial Intelligence: From Supervised to Semi-Supervised and Ultimately Unsupervised Technology Trends

The progression of Artificial Intelligence (AI) technology from supervised learning to semi-supervised methods and ultimately reaching the realm of unsupervised AI marks a remarkable evolution in the field. This article explores this captivating journey, tracing the development of AI from its roots in supervised learning, where models are trained using labeled data, to the innovative semi-supervised approach, which harnesses the power labeled and unlabeled data. The pinnacle of this evolution is unsupervised learning, where AI systems autonomously uncover hidden patterns and relationships within unlabeled data. The implications of this evolution are profound, shaping industries and sparking ethical conversations. This article delves into each stage, revealing the mechanics, applications, and potential societal impact of AI’s transformative trajectory. As we peer into the future, we anticipate an era of AI innovation characterized by unprecedented adaptability, creativity, and discovery.

Robust tool condition monitoring in Ti6Al4V milling based on specific force coefficients and growing self-organizing maps

Tool condition monitoring (TCM) is a mean to optimize production systems trying to use cutting tool life at its best. Nevertheless, nowadays available TCM algorithms typically lack robustness in order to be consistently applied in industrial scenarios. In this paper, an unsupervised artificial intelligence technique, based on Growing Self-Organizing Maps (GSOM), is presented in synergy with real-time specific force coefficients (SFC) estimation through the regression of instantaneous cutting forces. The conceived approach allows robustly mapping the SFC, exploiting process parameters and similarity to manage the variability of their estimation due to unmodelled phenomena, like machine dynamics and tool run-out. The devised approach allowed detecting the tool end-of-life in cutting tests with variable lubrication, machine tool and cutting speed, through the adoption of a self-starting control chart running on real-time clustered data. The solution was validated through the comparison of the GSOM framework with respect to the optimized self-starting control chart applied without GSOM clustering. The GSOM reached a root mean squared percentage error (RMSPE) of 13.2% with respect to 56.1% obtained with the analogous control chart in a full-set optimization scenario. When optimised on tests for a unique machine tool and tested on another machine tool, GSOM scored an RMSPE of 34.5%, whereas the optimized control chart scored 64.5%.

Patient Perspectives on Artificial Intelligence in Radiology

There are two major areas for patient engagement in radiology artificial intelligence (AI). One is in the sharing of data for AI development; the second is the use of AI in patient care. In general, individuals support sharing deidentified data if used for the common good, to help others with similar health conditions, or for research. However, there is concern with risk to privacy including reidentification and use for other than intended purposes. Lack of trust is mentioned as a barrier for data sharing. Individuals want to be involved in the data-sharing process. In the use of AI in medical care, patients generally support AI as an assist to the radiologist but lack trust in unsupervised AI. Patients worry about liability in case of bad outcomes. Patients are concerned about loss of the human connection and the loss of empathy during a vulnerable time in their lives. Patients expressed concern about risk of discrimination due to bias in AI algorithms. Building trust in AI requires transparency, explainability, security, and privacy protection. Radiologists can take action to prepare their patients to become more trusting of AI. Developing and implementing data-sharing agreements allows patients to voluntarily help in the algorithm development process. Developing AI disclosure guidelines and having AI use disclosure discussions with patients will help them understand the use of AI in their care. As the use of AI increases, there is an opportunity for radiologists to develop and maintain close relationships with their patients and to become more involved in their care.

Applications of artificial intelligence and bioinformatics methodologies in the analysis of ocular biofluid markers: a scoping review.

This scoping review summarizes the applications of artificial intelligence (AI) and bioinformatics methodologies in analysis of ocular biofluid markers. The secondary objective was to explore supervised and unsupervised AI techniques and their predictive accuracies. We also evaluate the integration of bioinformatics with AI tools. This scoping review was conducted across five electronic databases including EMBASE, Medline, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Web of Science from inception to July 14, 2021. Studies pertaining to biofluid marker analysis using AI or bioinformatics were included. A total of 10,262 articles were retrieved from all databases and 177 studies met the inclusion criteria. The most commonly studied ocular diseases were diabetic eye diseases, with 50 papers (28%), while glaucoma was explored in 25 studies (14%), age-related macular degeneration in 20 (11%), dry eye disease in 10 (6%), and uveitis in 9 (5%). Supervised learning was used in 91 papers (51%), unsupervised AI in 83 (46%), and bioinformatics in 85 (48%). Ninety-eight papers (55%) used more than one class of AI (e.g. > 1 of supervised, unsupervised, bioinformatics, or statistical techniques), while 79 (45%) used only one. Supervised learning techniques were often used to predict disease status or prognosis, and demonstrated strong accuracy. Unsupervised AI algorithms were used to bolster the accuracy of other algorithms, identify molecularly distinct subgroups, or cluster cases into distinct subgroups that are useful for prediction of the disease course. Finally, bioinformatic tools were used to translate complex biomarker profiles or findings into interpretable data. AI analysis of biofluid markers displayed diagnostic accuracy, provided insight into mechanisms of molecular etiologies, and had the ability to provide individualized targeted therapeutic treatment for patients. Given the progression of AI towards use in both research and the clinic, ophthalmologists should be broadly aware of the commonly used algorithms and their applications. Future research may be aimed at validating algorithms and integrating them in clinical practice.

Whole slide images-based prediction of lymph node metastasis in T1 colorectal cancer using unsupervised artificial intelligence.

Lymph node metastasis (LNM) prediction for T1 colorectal cancer (CRC) is critical for determining the need for surgery after endoscopic resection because LNM occurs in 10%. We aimed to develop a novel artificial intelligence (AI) system using whole slide images (WSIs) to predict LNM. We conducted a retrospective single center study. To train and test the AI model, we included LNM status-confirmed T1 and T2 CRC between April 2001 and October 2021. These lesions were divided into two cohorts: training (T1 and T2) and testing (T1). WSIs were cropped into small patches and clustered by unsupervised K-means. The percentage of patches belonging to each cluster was calculated from each WSI. Each cluster's percentage, sex, and tumor location were extracted and learned using the random forest algorithm. We calculated the areas under the receiver operator characteristics curves (AUCs) to identify the LNM and the rate of over-surgery of the AI model and the guidelines. The training cohort contained 217 T1 and 268 T2 CRCs, while 100 T1 cases (LNM-positivity 15%) were the test cohort. The AUC of the AI system for the test cohort was 0.74 (95% confidence interval [CI], 0.58-0.86), and 0.52 (95% CI, 0.50-0.55) using the guidelines criteria (p=0.0028). This AI model could reduce the 21% of over-surgery compared to the guidelines. We developed a pathologist-independent predictive model for LNM in T1 CRC using WSI for determination of the need for surgery after endoscopic resection.

A framework for fault detection and diagnostics of articulated collaborative robots based on hybrid series modelling of Artificial Intelligence algorithms

AbstractSmart factories build on cyber-physical systems as one of the most promising technological concepts. Within smart factories, condition-based and predictive maintenance are key solutions to improve competitiveness by reducing downtimes and increasing the overall equipment effectiveness. Besides, the growing interest towards operation flexibility has pushed companies to introduce novel solutions on the shop floor, leading to install cobots for advanced human-machine collaboration. Despite their reliability, also cobots are subjected to degradation and functional failures may influence their operation, leading to anomalous trajectories. In this context, the literature shows gaps in what concerns a systematic adoption of condition-based and predictive maintenance to monitor and predict the health state of cobots to finally assure their expected performance. This work proposes an approach that leverages on a framework for fault detection and diagnostics of cobots inspired by the Prognostics and Health Management process as a guideline. The goal is to habilitate first-level maintenance, which aims at informing the operator about anomalous trajectories. The framework is enabled by a modular structure consisting of hybrid series modelling of unsupervised Artificial Intelligence algorithms, and it is assessed by inducing three functional failures in a 7-axis collaborative robot used for pick and place operations. The framework demonstrates the capability to accommodate and handle different trajectories while notifying the unhealthy state of cobots. Thanks to its structure, the framework is open to testing and comparing more algorithms in future research to identify the best-in-class in each of the proposed steps given the operational context on the shop floor.

Artificial intelligence-based approach for cluster identification in a CFB riser

Cluster evolution dominates the gas–solid interaction and heat and mass transfer in risers, but cluster identification is still a challenge. Based on the unsupervised artificial intelligence approach, a two-step K-means cluster identification method is proposed to distinguish the clusters in a riser. With the consideration of the instantaneous particle motion and position information, it is able to capture the phenomenon of particles attaching and detaching a cluster. The estimated solid fraction threshold is located in the region where the slope of the solid fraction is the smallest, and approximately makes the numbers of particles detaching and attaching the clusters equal. The solid fraction threshold is about 0.204. It is found that the particles penetrating a cluster are in a more violent turbulent movement. Larger gas velocity and solid flux result in stronger interactions between clusters. The proposed method provides a new way to identify the cluster.

Advanced Control Systems in Industry 5.0 Enabling Process Mining.

This paper merges new research topics in Industry 5.0 using the Business Process Modeling and Notation (BPMN) approach able to integrate Artificial Intelligence (AI) in production processes. The goal is to provide an innovative approach to model production management in industry, adopting a new "proof of concept" of advanced Process Mining (PM) automatizing decisions and optimizing machine setting and maintenance interventions. Advanced electronic sensing and actuation systems, integrating supervised and unsupervised AI algorithms, are embedded in the PM model as theoretical process workflows suggested by a Decision Support System (DSS) engine enabling an intelligent decision-making procedure. The paper discusses, as examples, two theoretical models applied to specific industry sectors, such as food processing and energy production. The proposed work provides important elements of engineering management related to the digitalization of production process matching with automated control systems setting production parameters, thus enabling the self-adapting of product quality supervision and production efficiency in modern industrial systems.

Assessing Surface Shapes of the Optic Nerve Head and Peripapillary Retinal Nerve Fiber Layer in Glaucoma with Artificial Intelligence.

PurposeTo assess 3-dimensional surface shape patterns of the optic nerve head (ONH) and peripapillary retinal nerve fiber layer (RNFL) in glaucoma with unsupervised artificial intelligence (AI).DesignRetrospective study.ParticipantsPatients with OCT scans obtained between 2016 and 2020 from Massachusetts Eye and Ear.MethodsThe first reliable Cirrus (Carl Zeiss Meditec, Inc) ONH OCT scans from each eye were selected. The ONH and RNFL surface shape was represented by the vertical positions of the inner limiting membrane (ILM) relative to the lowest ILM vertical position in each eye. Nonnegative matrix factorization was applied to determine the ONH and RNFL surface shape patterns, which then were correlated with OCT and visual field (VF) loss parameters and subsequent VF loss rate. We tested whether using ONH and RNFL surface shape patterns improved the prediction accuracy for associated VF loss and subsequent VF loss rates measured by adjusted r2 and Bayesian information criterion (BIC) difference compared with using established OCT parameters alone.Main Outcome MeasuresOptic nerve head and RNFL surface shape patterns and prediction of the associated VF loss and subsequent VF loss rates.ResultsWe determined 14 ONH and RNFL surface shape patterns using 9854 OCT scans from 5912 participants. Worse mean deviation (MD) was most correlated (r = 0.29 and r = 0.24, Pearson correlation; each P < 0.001) with lower coefficients of patterns 10 and 12 representing inferior and superior para-ONH nerve thinning, respectively. Worse MD was associated most with higher coefficients of patterns 5, 4, and 9 (r = –0.16, r = –0.13, and r = –0.13, respectively), representing higher peripheral ONH and RNFL surfaces. In addition to established ONH summary parameters and 12–clock-hour RNFL thickness, using ONH and RNFL surface patterns improved (BIC decrease: 182, 144, and 101, respectively; BIC decrease ≥ 6; strong model improvement) the prediction of accompanied MD (r2 from 0.32 to 0.37), superior (r2 from 0.27 to 0.31), and inferior (r2 from 0.17 to 0.21) paracentral loss and improved (BIC decrease: 8 and 8, respectively) the prediction of subsequent VF MD loss rates (r2 from 0 to 0.13) and inferior paracentral loss rates (r2 from 0 to 0.16).ConclusionsThe ONH and RNFL surface shape patterns quantified by unsupervised AI techniques improved the structure–function relationship and subsequent VF loss rate prediction.

Research on Positioning and Monitoring Method of Pipeline Abnormal State Based on Sliding Window Outlier Analysis

In recent years, various accidents caused by pipeline corrosion and groundwater infiltration have brought irreparable losses to people’s lives and property safety. Therefore, it is very important to lay sensors on the pipeline for real-time monitoring and alarm. Aiming at the problem of distributed optical fiber sensing in the real-time pipeline monitoring process, this study proposes a pipeline invasion detection analysis method based on sliding window outliers. By continuously optimizing the value of the sliding window and the length of the abnormal state diagnosis window, the false alarm rate of the system is greatly reduced. This method only uses amplitude data to realize the intelligent identification of abnormal pipeline conditions. In addition, a simulation experiment of pipeline groundwater invasion monitoring has been carried out, and the results show that the algorithm does not generate false positives when the device is in good condition. Once the pipeline is invaded, the algorithm can quickly determine the location, and the error range is stable at 0.5 m. This method is an unsupervised artificial intelligence pipeline invasion detection processing method, and it has a good application prospect in the monitoring of abnormal pipeline conditions.

A new linear combination method of haplogroup distribution central vectors to model population admixtures

We introduce a novel population genetic approach suitable to model the origin and relationships of populations, using new computation methods analyzing Hg frequency distributions. Hgs were selected into groups which show correlated frequencies in subsets of populations, based on the assumption that correlations were established in ancient separation, migration and admixture processes. Populations are defined with this universal Hg database, then using unsupervised artificial intelligence, central vectors (CVs) are determined from local condensations of the Hg-distribution vectors in the multidimensional point system. Populations are clustered according to their proximity to CVs. We show that CVs can be regarded as approximations of ancient populations and real populations can be modeled as weighted linear combinations of the CVs using a new linear combination algorithm based on a gradient search for the weights. The efficacy of the method is demonstrated by comparing Copper Age populations of the Carpathian Basin to Middle Age ones and modern Hungarians. Our analysis reveals significant population continuity since the Middle Ages, and the presence of a substrate component since the Copper Age.

Unsupervised explainable AI for molecular evolutionary study of forty thousand SARS-CoV-2 genomes

BackgroundUnsupervised AI (artificial intelligence) can obtain novel knowledge from big data without particular models or prior knowledge and is highly desirable for unveiling hidden features in big data. SARS-CoV-2 poses a serious threat to public health and one important issue in characterizing this fast-evolving virus is to elucidate various aspects of their genome sequence changes. We previously established unsupervised AI, a BLSOM (batch-learning SOM), which can analyze five million genomic sequences simultaneously. The present study applied the BLSOM to the oligonucleotide compositions of forty thousand SARS-CoV-2 genomes.ResultsWhile only the oligonucleotide composition was given, the obtained clusters of genomes corresponded primarily to known main clades and internal divisions in the main clades. Since the BLSOM is explainable AI, it reveals which features of the oligonucleotide composition are responsible for clade clustering. Additionally, BLSOM also provided information concerning the special genomic region possibly undergoing RNA modifications.ConclusionsThe BLSOM has powerful image display capabilities and enables efficient knowledge discovery about viral evolutionary processes, and it can complement phylogenetic methods based on sequence alignment.

On crack tip localisation in quasi-statically loaded, adhesively bonded double cantilever beam specimens by acoustic emission

The feasibility of acoustic emission structural health monitoring to detect, localise and monitor crack propagation during quasi-static mode I loading of adhesively bonded joints was studied. Unsupervised artificial intelligence pattern recognition methods (Self-Organised maps and K-means) were used to classify acoustic emission raw data as either background noise or relevant information. After that, three different time-of-arrival picking algorithms were considered and implemented to determine the acoustic emission source’s location, and their accuracy was discussed. Localised acoustic emission events were divided into well-defined groups with different energy levels and compared to Digital Image Correlation and visual evaluation results. It was possible to conclude that the highest energetic group allows the assessment of the onset of plasticisation ahead of the crack-tip within the studied adhesive, bringing novel standpoints to the use of acoustic emission as a structural health monitoring method for adhesively bonded joints.

Breaking the ageing paradigm in endometrium: endometrial gene expression related to cilia and ageing hallmarks in women over 35 years

AbstractSTUDY QUESTIONDoes age affect endometrial gene expression?SUMMARY ANSWERUsing unsupervised artificial intelligence methods, we report for the first time that endometrial gene expression changes from 35 years of age in women.WHAT IS KNOWN ALREADYFemale fertility declines with age, largely attributed to declining oocyte quality and ovarian reserve. Combined with other evidence, a longstanding paradigm holds that age does not affect the endometrial function and age has not been controlled for properly in endometrial studies.STUDY DESIGN, SIZE, DURATIONA retrospective in silico analysis was performed of endometrial transcriptomic data from the Gene Expression Omnibus (GEO) sample repository for 27 women of different ages. Results were validated in an independent gene expression dataset of 20 endometrial samples from women aged 23–43 years.PARTICIPANTS/MATERIALS, SETTING, METHODSA systematic search was performed in GEO from October 2016 to January 2019 to identify transcriptomic studies involving women of different ages. Included samples were from norm-ovulatory, women of reproductive age (23–49 years) with regular menstrual cycles who were free of endometriosis and used as controls in a previous endometrial study. We used raw gene expression data and metadata from these samples to investigate the effect of age on endometrial gene expression. Files were downloaded, pre-processed and explored for potential confounding variables and outliers. Artificial intelligence methods were applied to define age groups, and differential expression and functional analyses were applied to demonstrate and understand the effect of age on gene expression at the molecular level. Functional results were validated in an independent gene expression dataset of 20 endometrial samples from women aged 23–43 years.MAIN RESULTS AND THE ROLE OF CHANCEAnalysis of the initially retrieved endometrial datasets revealed the age of participants was not available (33.33%) or traceable (43.33%) in most studies. However, one study was suitable for age analysis (GSE4888, n = 27, 23–49 years). Samples showed different transcriptomic profiles according to age, beginning at 35 years. A total of 5778 differentially expressed genes and 27 significantly altered endometrial functions (false discovery rate (FDR) &amp;lt; 0.05) were associated with endometrial gene expression changes related to age. Interestingly, 81.48% of affected functions were related to up-regulation of ciliary processes, with 91 genes involved in cilia motility and ciliogenesis. Other functions included dysregulation of the vascular endothelial growth factor signalling pathway and inhibition of epithelial proliferation triggered by 37 genes involved in cell cycle arrest, angiogenesis, insulin signalling and telomere protection. These findings were validated in an independent dataset using a non-targeted approach; 20 up-regulated ciliary processes (FDR &amp;lt; 0.02) and 6 down-regulated functions related to cell cycle arrest were identified as affected by age, among other hallmarks of ageing such as DNA repair inhibition or sugar metabolism (FDR &amp;lt; 0.05).LARGE SCALE DATAData underlying this article are available in GEO, IDs: GSE4888 (main dataset) and GSE102131 (validation dataset).LIMITATIONS, REASONS FOR CAUTIONThis study is limited in size, as are most studies of endometrial transcriptomics where whole-transcriptome analysis considers nearly 22 000 variables in a relatively small population. Yet, our study includes a main sample set and subsequent validation set that enhances reproducibility of our results and provides reasonable evidence for concluding that age affects endometrial gene expression. A larger study prospectively controlling for patient characteristics is needed to accurately describe changes related to age, with a higher sample size and across a wide age range. Additional studies also are necessary to determine the endometrial ageing contribution to infertility for ultimate translation to a clinical setting.WIDER IMPLICATIONS OF THE FINDINGSOur findings support an influence of age on the endometrium in a genome-wide functional approach, breaking the endometrial ageing paradigm in human reproduction. To our knowledge, this work is the first to identify, using a genome-wide functional non-targeted approach, ciliary processes as the primary dysregulated function associated with maternal age. These results should guide the research community to control for age as a potential confounding variable in endometrial gene expression studies and to consider endometrial ageing in further studies as a potential cause of infertility in the clinical setting. The reported functional dysregulations could contribute to diminished embryo implantation with age and further studies will demonstrate if such dysregulation underlies some cases of implantation failure. Additionally, the discovery of these functional alterations could enable mechanistic studies, particularly around the age-related increase in uterine pathologies.STUDY FUNDING/COMPETING INTEREST(S)This research was funded by the Instituto de Salud Carlos III through Miguel Servet programme (CP20/00118) granted to Patricia Diaz-Gimeno (Spanish Government) co-funded by FEDER; and by IVI Foundation (1706-FIVI-041-PD). A.D.-P. (FPU/15/01398) and A.P.-L. (FPU18/01777) are granted by the pre-doctoral programme fellowship from the Ministry of Science, Innovation and Universities (Spanish Government). The authors do not have any competing interests to declare.TRIAL REGISTRATION NUMBERN/A

Whole Slide Images-Based Prediction of Lymph Node Metastasis in T1 Colorectal Cancer Using Unsupervised Artificial Intelligence

Whole Slide Images-Based Prediction of Lymph Node Metastasis in T1 Colorectal Cancer Using Unsupervised Artificial Intelligence

AI-Driven Clinical Decision Support: Enhancing Disease Diagnosis Exploiting Patients Similarity

Detecting diseases at early stage can help to overcome and treat them accurately. A Clinical Decision Support System (CDS) facilitates the identification of diseases together with the most suitable treatments. In this paper, we propose a CDS framework able to integrate heterogeneous health data from different sources, such as laboratory test results, basic information of patients, health records and social media data. Using the data so collected, innovative machine learning and deep learning approaches can be employed. A neural network model for predicting patients&#x2019; future health conditions is proposed. The approach employs word embedding to model the semantic relations of hospital admissions, symptoms and diagnosis, and it introduces a mechanism to measure the relationships of different diagnosis in terms of symptoms similarity to exploit for the prediction task. Several CDSs, including diagnostic decision support systems for inferring patient diagnosis, have been proposed in the literature. However, these methods typically focus on a single patient and apply manually or automatically constructed decision rules to produce a diagnosis. Even worst, they consider only a single medical condition, whereas it is not uncommon that a patient has more than one medical condition at the same time. The novelty of the proposed approach is the combination of supervised and unsupervised artificial intelligence methods allowing to combine several and heterogeneous data sources related to a multitude of patients and concerning different medical conditions. Furthermore, with respect to previous approaches, the diagnosis prediction problem is formulated to predict the exact diagnosis in terms of semantic meaning by exploiting Natural Language Processing concepts. Experimental results, performed on a real-world EHR dataset, show that the proposed approach is effective and accurate and provides clinically meaningful interpretations. The obtained outcomes are promising for future extensions of the framework that could be a valuable means for automatic inferring disease diagnosis.

Radiogenomics and Artificial Intelligence Approaches Applied to Cardiac Computed Tomography Angiography and Cardiac Magnetic Resonance for Precision Medicine in Coronary Heart Disease: A Systematic Review.

The risk of coronary heart disease (CHD) clinical manifestations and patient management is estimated according to risk scores accounting multifactorial risk factors, thus failing to cover the individual cardiovascular risk. Technological improvements in the field of medical imaging, in particular, in cardiac computed tomography angiography and cardiac magnetic resonance protocols, laid the development of radiogenomics. Radiogenomics aims to integrate a huge number of imaging features and molecular profiles to identify optimal radiomic/biomarker signatures. In addition, supervised and unsupervised artificial intelligence algorithms have the potential to combine different layers of data (imaging parameters and features, clinical variables and biomarkers) and elaborate complex and specific CHD risk models allowing more accurate diagnosis and reliable prognosis prediction. Literature from the past 5 years was systematically collected from PubMed and Scopus databases, and 60 studies were selected. We speculated the applicability of radiogenomics and artificial intelligence through the application of machine learning algorithms to identify CHD and characterize atherosclerotic lesions and myocardial abnormalities. Radiomic features extracted by cardiac computed tomography angiography and cardiac magnetic resonance showed good diagnostic accuracy for the identification of coronary plaques and myocardium structure; on the other hand, few studies exploited radiogenomics integration, thus suggesting further research efforts in this field. Cardiac computed tomography angiography resulted the most used noninvasive imaging modality for artificial intelligence applications. Several studies provided high performance for CHD diagnosis, classification, and prognostic assessment even though several efforts are still needed to validate and standardize algorithms for CHD patient routine according to good medical practice.